A deal desk that reads the internet every six hours.
Click Frenzy’s team used to find deals by reading retailer sites and promo emails by hand; FrenzyOS now re-ranks every deal on a 1-hour cycle and crawls retailer sites every 6 hours.

Click Frenzy runs major Australian online shopping events, and its team needs to know which deals are worth featuring. We built FrenzyOS, which finds deals across retailer sites and promo emails, reads them with AI and ranks them.
AI level 2 to 3. What leaving Level 2 takes
What it uses: Retailer sites, Promo emails, AI extraction, Ranking models, Deal chatbot
Visit Click FrenzyFrom inbox to shortlist
A promo email becomes a ranked deal.
Every six hours the crawlers visit retailer sites, and the deals inbox is checked every hour. AI reads each page or email and pulls out the deal itself. Each one is enriched, scored on demand, value and the store’s reputation, and re-ranked every hour, so the team can simply ask what’s worth featuring.
Retailer sites crawled
Every six hours.
Promo emails picked up
Every hour.
AI reads each one and extracts the deal
Scored on demand, value and store reputation
With models trained on past deal data.
Re-ranked hourly, ready to ask
In the admin views or the chatbot.
The problem
Too many deals, not enough eyes.
Australian retailers publish deals constantly, across their own sites and their mailing lists. Finding the good ones by hand meant someone reading all of it, all the time.
Deals everywhere
Spread across retailer sites and inboxes.
Value is hard to judge quickly
A big discount isn’t always a good deal.
Found too late
Deals move faster than a manual check.
How often
- 6 hrs
- between full crawls of retailer sites
- 1 hr
- between re-ranks of every deal
- 3
- signals in every score: demand, value and store reputation
- 1
- chatbot to ask what’s worth featuring
How it’s put together
The judgement is Click Frenzy’s. The legwork is automatic.
Crawlers, email intake, extraction, scoring and the chatbot were built for Click Frenzy. The scoring models learn from past deal data and search trends, and the team still decides what gets featured.
Connects to
- Retailer sites
- Promo inbox
- Search trends
- Deal community data
Built for Click Frenzy
- Crawlers and email intake
- AI deal extraction
- Demand and value scoring
- Deal chatbot and admin views
Built on Rome, the enterprise operating system for AI.
Your deal research
How much reading is your team doing?
How the estimate works
Hours a week = sources × checks a week × minutes per check ÷ 60.
No fixed assumptions. It uses only your three answers.
The example figures are illustrative. Your estimate is separate from the results shown for Click Frenzy, and it is not a measured saving or a guarantee.
The playbook, free
Build your own AI agent for one weekly job
Set up an agent that prepares one recurring job for a person to approve, with the AI tools you already pay for.